An Introduction to EMC CLARiiON Storage Device Technology

Applied Technology

Abstract This white paper compares and contrasts the merits of four types of storage devices that can be used on Dell/EMC CX4 and Dell AX4-5 stor-age arrays: enterprise flash drive (EFD), Fibre Channel (FC), Serial Attach SCSI (SAS), and Serial ATA (SATA).

Dell Inc. April 2009 Visit www.Dell.com/emc for more information on Dell/EMC Storage. Copyright © 2009 Dell Inc. THIS WHITE PAPER IS FOR INFORMA-TIONAL PURPOSES ONLY, AND MAY CONTAIN TYPOGRAPHICAL ER-RORS AND TECHNICAL INACCURACIES. THE CONTENT IS PROVIDED AS IS, WITHOUT EXPRESS OR IMPLIED WARRANTIES OF ANY KIND.

An Introduction to

Dell/EMC Storage

Device Technology

An Introduction to Dell/EMC Storage Device Technology

Applied Technology 2

Table of Contents

Executive Summary ............................................................................................3

Introduction .........................................................................................................3

Audience ....................................................................................................................................... 3

Terminology .................................................................................................................................. 4

Enterprise Flash Drive Technology ...................................................................5

Wear Leveling ............................................................................................................................... 5

Multiple I/O Channels ................................................................................................................... 5

Flash cell architecture................................................................................................................... 6

Hard Drive Technology .......................................................................................7

The Logical Drive ................................................................................................8

Enterprise Flash Drives ......................................................................................8

Dell/EMC 73 GB (4 Gb/s) EFDs ................................................................................................... 8

Performance of 73 GB EFDs .................................................................................................... 9 Applications using 73 GB EFDs ................................................................................................ 9

Fibre Channel Hard Drives ............................................................................... 10

Dell/EMC 146, 300, and 450 GB 15k rpm (4 Gb/s) FC drives ................................................... 10

Performance of 15K RPM Disk Drives .................................................................................... 10 Applications Using 146, 300, and 450 GB 15K RPM FC Drives ............................................ 10

Dell/EMC 300 and 400 GB 10K RPM (4 Gb/s) FC Drives ......................................................... 10

Performance of 10K RPM Disk Drives .................................................................................... 11 Applications Using 300 and 400 GB 10K RPM Disk Drives ................................................... 11

Serial Attach SCSI Hard Drives ........................................................................ 12

Dell AX4-5 400 GB 10K and 15K RPM SAS Disk Drives ........................................................... 12

Serial ATA Hard Drives ..................................................................................... 12

Dell/EMC 1 TB 7200 RPM SATA II Drives ................................................................................. 12

Benefits of 1 TB SATA II Disk Drive Technology .................................................................... 13 Dell/EMC 1 TB 5400 RPM SATA II Drives (Lower Power SATA) .............................................. 13

Implementing 1 TB SATA II Disk Drives ................................................................................. 13 SATA II "Northstar" Disk Drive and Enclosure Technology ....................................................... 14

Disk Drive Performance Comparisons ............................................................ 14

Competitive Advantages of the Dell/EMC with ATA ................................................................... 15

Implementing Various Technology Dell/EMC Disk Drives ............................. 15

A Few Examples of Mixed Disk Drive Usage ............................................................................. 15

Dell/EMC Disk Drive Power Solutions ........................................................................................ 16

Multi-Tiering ................................................................................................................................ 16

Rebuild Times ......................................................................................................................... 17 Performance Planning ............................................................................................................ 18

Dell Hard Drive Reliability Qualification .......................................................... 18

Conclusion ......................................................................................................... 19



Executive Summary There are four types of storage device technologies available for use on Dell/EMC CX4 and Dell AX4-5 storage

arrays: Enterprise Flash Drive (EFD), Fibre Channel (FC), Serial Attach SCSI (SAS), and Serial ATA (SATA).

While all of these drives store data, they have different operating characteristics, interfaces, and price points that

make them more suitable to different operating environments. For example FC, SAS, and SATA drives have

spinning media whereas enterprise flash drives have solid state media – this characteristic makes them suitable for

different kind of environments.

There are two main types of operating environments: performance-intensive environments and capacity-intensive

environments. Performance environments need high throughput, high bandwidth, and high reliability. Typical

performance environments are transaction-based systems with workloads characterized by random reads and writes.

Capacity-intensive environments need large amounts of low-cost storage, high reliability, and modest throughput

and bandwidth performance. Typical capacity-intensive environments are archival systems with a workload

characterized by sequential reads.

EFDs are recommended for performance-intensive applications or parts of applications with low-response-time and

high-throughput requirements. FC hard drives should continue to be the choice for environments with large-capacity

high-performance requirements. SAS hard drives provide performance and reliability that are equivalent to FC

drives. SATA hard drives are the choice in modest-performance and high-capacity environments. In addition, SATA

drives can provide energy-efficient bulk storage capacity at low cost.

However, the choice between performance and capacity is not always clear. Many Dell/EMC systems operate in

―mixed‖ environments that have a primary mission to provide storage for a performance workload but must also

provide storage for a capacity-intensive workload. In this case, a blend of storage device types (EFD, FC, and

SATA on CX arrays or SAS and SATA on AX arrays) may be the most appropriate and cost-effective solution. In

addition, there are Dell/EMC systems that operate in environments with requirements for both performance and

capacity. In this case, a high-capacity ―EFD and FC‖ or ―all SAS‖ drive solution is merited. Choice of a drive type

also depends on the type of Dell/EMC storage system. Dell/EMC CX arrays support EFD, FC, and SATA drive

types while Dell AX arrays support SAS and SATA drives only.

Introduction This white paper compares and contrasts the merits of the different drive types (EFD, FC, SAS, and SATA) that can be

used on the Dell/EMC CX4 and Dell AX4-5 storage arrays.

With more open-system storage capacity shipping this year than in all past years combined, a new standard for

storage system performance is required to access all this data rapidly. With its Dell/EMC storage systems, Dell steps

up to meet this challenge with its lineup of some of the fastest and highest capacity storage devices available on the

market today. The 1 TB 5400/7200 rpm SATA II drives, 400 GB 10k rpm Fibre Channel/SAS drives, 146/300/450

GB 15k rpm Fibre Channel/SAS drives, and 73 GB enterprise flash drives all help redefine storage-system

capacities and performance. Larger capacity drives continue to be released over time, offering more options within

each technology family.

Audience This white paper is intended for Dell sales engineers, professional services, and customers wanting an introduction

on the types of storage devices and their characteristics available on the Dell/EMC CX4 and Dell AX4-5 storage

arrays.



Terminology Average seek time ― The average amount of time it takes for the head stack assembly to get from an original

random point on the media surface (A) to a random destination point on the media surface (B). Seeking involves

mechanical motion of the drive head and takes considerable time (relative to other non-mechanical, electronic

operations inside a storage array. Average seek time is obtained by averaging the seek times for a given number of

random seeks over a certain period of time. Naturally, this time is reduced as the spindle motor speed is increased

because data becomes available more quickly. Seek times are measured separately for read and write operations.

Read operations take less time than write operations.

Flash memory – flash memory is a non-volatile memory that can be electrically erased and reprogrammed. It is a

specific type of electrically erasable programmable read-only memory (EEPROM) that is erased and programmed in

blocks. It does not need any power to maintain the information stored in the chip. The individual cells used for

storing bits of information can be of NOR or NAND architecture.

Fibre Channel ― A high-bandwidth data-transfer technology that uses optical cables to connect devices and is

widely used to connect servers to high-performance storage systems in a SAN.

Magic 8 bytes – All Dell/EMC systems use a disk sector format of 520 bytes. This format allows the storage of 512

bytes of data, as in most disk systems, but also provides 8 bytes per sector for metadata, which EMC® algorithms

use for data integrity checks. These additional bytes include a linear check sum, write stamp, parity shed stamp, and

time stamp. The RAID protection level of the sector determines which of these stamps is used by the array software.

Rotational latency ― When the head arm moves to a track to access data, it must wait for the right sector to appear

under the head. This takes time. On average, the platter must rotate half of a revolution to reach the sector. This

latency time is inversely proportional to the disk rotation speed and is reduced as the spindle motor speed increases.

The average rotational latency specification of a 15,000 rpm drive is 2.0 ms, or typically half the time it takes to

complete one full revolution of the disk surface itself.

Serial Advanced Technology Attachment ( SATA) ― An enhancement that uses thinner cables and provides

better performance over parallel drive technology. SATA II includes improvements such as Native Command

Queuing (NCQ) to improve performance and reliability.

Serial Attach SCSI (SAS) ― An I/O technology that uses a serial point-to-point interface instead of a parallel bus

interface, as with parallel SCSI.

Spindle motor speed ― The actual speed of the spindle motor assembly measured in rpm or revolutions per

minute. One complete revolution is the amount of time it takes to make one full revolution from a given start point

on the disk surface back to the given start point. The amount of time it takes on a 15,000 rpm disk drive to make one

full revolution is 4 ms, compared to approximately 6 ms on a 10,000 rpm disk drive. This is a roughly 33 percent

decrease in the amount of time it takes to make one full revolution of the disk media surface.

Transfer rate ― The rate at which data is transferred from either the drive to the target host (external) or from the

media surface to the head assembly (internal). Although the internal transfer rate of the hard disk increases as the

spindle speed increases (head-to-media transfer rate), the external transfer rate typically stays the same due to the

interface chipset architecture of the hard disk drive itself. Unless a higher transfer rate interface chipset is used on

the hard disk drive itself, the external transfer rate will stay the same as that of the predecessor drive.



Enterprise Flash Drive Technology EFDs, commonly known in the IT world as solid state drives (SSDs), that are used in Dell/EMC CX4 storage arrays

differ significantly from the solid state technology used in consumer electronics, particularly in their performance

and reliability characteristics. Consumer electronics are built around commodity class technology where price is the

primary consideration. These consumer grade devices do not have the performance and resiliency that enterprise

applications require. EFDs address the performance and reliability challenges that are associated with flash

technology. A Dell enterprise flash drive is a sophisticated system in itself that has been designed to meet the needs

of enterprise customers and their applications. EFDs have dual-ported front-end controller interfaces that allow the

drive to deliver data over two different channels; a highly protected and integrated processing engine; a large

amount of DRAM to act as a buffer between the array and the slower flash media; and the ability to simultaneously

process multiple read and write I/Os to the flash media in the back end. The drive is also protected end to end with

error detection/correction technology to detect and correct bit errors that may occur. There is a high level of

redundancy and error correction built into each drive to make sure that data corruption is always detected and

corrected.

EFDs are especially well suited for low-latency applications that require consistent, low (less than 1 ms) read/write

response times. The greatest improvements will be seen with highly concurrent, random read workloads, owing to

the lack of rotational and seek latency in EFDs. The elimination of mechanical overhead and data placement latency

greatly improves application performance and efficiency. All of these factors contribute to the fact that EFDs on

Dell/EMC arrays can deliver very high IOPS with very low response times without physical spindles.

Besides the industry standard techniques used in all Dell/EMC storage devices (like DRAM memory between the

array and the storage media; a backup power circuit for the DRAM memory; error detection and correction; and bad

block management), there are certain features that are specific to EFDs. These features are discussed in the

following sections.

Wear Leveling Individual flash cells support a finite number of erase-write cycles. Utilizing this technology in EFD enterprise

storage requires features that improve the drive’s reliability. EFDs have more raw NAND flash capacity internally

than the usable amount visible to the Dell/EMC array. Dell EFDs balance erase and rewrite operations across their

entire raw capacity to sustain maximum bandwidth and reliability, and to optimize the usable life of the drive. EFDs

employ this wear-leveling technique to ensure that all cells in the drive are used evenly, and to minimize the risk of

―wear-out‖ to which other industry flash drive cells are susceptible. This wear-leveling is performed separately for

each I/O channel within the EFD. There are two types of wear-leveling:

Dynamic wear leveling - Pointers to all available blocks within an EFD are maintained in a free list. When the

number of blocks in the free list drops below a minimum threshold, a background process frees up blocks with

the least amount of data by consolidating their data into a single block – similar to the de-fragmentation process

in many operating systems. The blocks that are freed up are erased and added to the free list. Since any new

block is always allocated from the free list, the drive firmware ensures that the wear is distributed evenly across

all available flash cells in the drive.

Static wear leveling - EFD firmware maintains a list of how often blocks have been erased, which is a measure

of how ―worn‖ a block is. With wear-leveling, the contents of blocks with high erase counts are moved to

blocks with the low erase counts. The drive firmware keeps the most erased block and least erased block on an

I/O channel within a set number of erase cycles of each other.

Multiple I/O Channels Dell enterprise flash drives have multiple internal independent I/O channels, which are also referred to as threads.

This is like having 16 independent service centers for the incoming I/O requests. This allows EFDs to service

multiple requests at the same time, unlike traditional spinning drives that can service only one request at a time.

This ability greatly enhances the performance of applications with higher concurrency (or parallelism).



Flash cell architecture In general, there are three major technologies used in all flash drives:

NOR-based flash cell

Single-Level Cell (SLC) NAND flash cell

Multi-Level Cell (MLC) NAND flash cell

NOR flash technology is primarily used where high-density memory and high performance are not the main

requirements. It is best suited for code storage and execution – usually in low densities. This makes it ideal for chips

that store program code that is rarely updated, such as ROM BIOS code in a motherboard.

NAND flash cells have a very compact architecture; their cell size is almost half the size of a comparable NOR cell.

This characteristic, when combined with a simpler production process, enables a NAND cell to offer higher

densities with more memory on a given die size. This results in a lower cost per gigabyte. With smaller, more

precise manufacturing processes being used in the future, their price is expected to fall even further. A NAND flash

cell also has faster erase and write times compared to NOR-based flash cells, thus providing improved performance.

It has higher endurance limits for each cell, a feature that provides the reliability required for enterprise-class

applications.

Flash storage devices store information in a collection of flash cells made from floating gate transistors. Single-

Level Cell (SLC) devices store only one bit of information in each cell (binary), whereas Multi-Level Cell (MLC)

devices store more than one bit per flash cell by choosing between multiple levels of electrical charge to apply to its

floating gates in the transistors (Figure 1).

Figure 1. Comparison between MLC and SLC flash cell data storage

Since each cell in MLC flash has more information bits, an MLC flash-based storage device offers increased storage

density compared to a SLC flash-based version. The downside is that MLC NAND has lower performance and

reliability because of its inherent architectural limitations. Higher functionality further complicates the use of MLC

NAND, making it necessary to implement more advanced flash management algorithms and controllers. SLC and



MLC NAND offer capabilities that serve two very different types of applications – those requiring high performance

at an attractive cost per bit (MLC) and those seeking even higher performance over time that are less cost sensitive

(SLC).

Taking into account the kind of I/O profiles in enterprise applications and their requirements, Dell enterprise flash

drives have the SLC NAND flash architecture. Table 1 compares the SLC and MLC flash characteristics.

Table 1. SLC and MLC flash comparison

Features MLC

Capacity

SLC Performance

Reliability

Bits per cell 2 1

Endurance (ERASE/WRITE cycles) <10K <100K

Read (Max) 50μs 25μs

Write (Typical) 600–900μs 200–300μs

Block Erase (Typical) 2 ms 1.5–2 ms

Although SLC NAND offers a lower density, it also provides an enhanced level of performance in the form of faster

reads and writes. Because SLC NAND stores only one bit per cell, the likelihood for error is reduced. SLC also

allows for higher write/erase cycle endurance, making it a better fit for use in applications requiring higher

reliability, and increased endurance and viability in multi-year product life cycles.

Hard Drive Technology

Figure 2. Physical drive

Hard disk drives (HDDs) record data by magnetizing a ferromagnetic material directionally, to represent either a 0

or a 1 binary digit. The drives read the data back by detecting the magnetization of the material. A typical HDD

design consists of a spindle that holds one or more flat circular disks called platters, onto which the data is recorded.

The platters are made from a non-magnetic material, usually glass or aluminum, and are coated with a thin layer of

magnetic material. Older disks used iron (III) oxide as the magnetic material, but current disks use a cobalt-based

alloy.

The platters are spun at very high speeds. Information is written to a platter as it rotates past mechanisms called

read-and-write heads that operate very close to the magnetic surface. The read-and-write head is used to detect and

modify the magnetization of the material immediately under it. There is one head for each magnetic platter surface

on the spindle, mounted on a common arm. An actuator arm (or access arm) moves the heads on an arc (roughly



radially) across the platters as they spin, allowing each head to access almost the entire surface of the platter as it

spins. The arm is moved using a voice coil actuator or (in older designs) a stepper motor.

The magnetic surface of each platter is divided into many small sub-micrometer size magnetic regions, each of

which is used to encode a single binary unit of information. In today's HDDs each of these magnetic regions is

composed of a few hundred magnetic grains. Each magnetic region forms a magnetic dipole that generates a highly

localized magnetic field nearby. The write head magnetizes a magnetic region by generating a strong local magnetic

field nearby. Early HDDs used an electromagnet both to generate this field and to read the data by using

electromagnetic induction. Later versions of inductive heads included Metal in Gap (MIG) heads and thin film

heads. In today's heads, the read and write elements are separate but in close proximity on the head portion of an

actuator arm. The read element is typically magneto-resistive while the write element is typically thin-film

inductive.

In modern drives, the small size of the magnetic regions creates the danger that their magnetic state will be lost

because of thermal effects. To counter this, the platters are coated with two parallel magnetic layers, separated by a

3-atom-thick layer of the non-magnetic element ruthenium, and the two layers are magnetized in opposite

orientation, thus reinforcing each other. Another technology used to overcome thermal effects to allow greater

recording densities is perpendicular recording, which has been used in some hard drives as of 2006.

Hard disk drives are sealed to prevent dust and other sources of contamination from interfering with the operation of

the hard disks heads. The hard drives are not air tight, but rather utilize an extremely fine air filter, to allow for air

inside the hard drive enclosure. The spinning of the disks causes the air to circulate, forcing any particulates to

become trapped on the filter. The same air currents also act as a gas bearing, which enables the heads to float on a

cushion of air above the surfaces of the disks.

The Logical Drive

A logical drive is a device that provides an area of usable storage capacity on one or more physical disk drive

components in a computer system. It is also referred to as logical volume or a Logical Unit Number (LUN), and in

some cases a virtual disk.

The disk is described as logical because it does not actually exist as a single physical entity in its own right. Logical

disks are also defined at various levels in the storage infrastructure stack. The levels, from top to bottom, are:

Operating system: Defines partitions on the disks to which it has visibility (these disks may be logical

themselves).

SAN: If the SAN is virtualized, a device in the same SAN presents logical drives to the host operating systems.

Storage subsystem: Usually providing some form of RAID where logical drives (partitions) are presented to the

SAN from the RAID arrays themselves. The RAID arrays actually contain the physical disks.

Enterprise Flash Drives

Dell/EMC 73 GB (4 Gb/s) EFDs With this capability, Dell/EMC has a Tier 0 ultra-performance storage tier that transcends the limitations previously

imposed by magnetic disk drives. For years, the most demanding enterprise applications have been limited by the

performance of magnetic rotating disk media. Tier 1 performance in storage arrays has been unable to surpass the

physical limitations of rotating disk drives. With Dell’s addition of EFDs to Dell/EMC CX4, organizations can now

take advantage of the drive’s ultra-high performance that is optimized for the highest level enterprise requirements.

EFDs contain no moving parts and appear as standard Fibre Channel drives to existing management tools, allowing

administrators to manage Tier 0 without special processes or custom tools. Tier 0 flash storage is ideally suited for

applications with high transaction rates and those requiring the fastest possible retrieval and storage of data.

http://en.wikipedia.org/wiki/Micrometre

http://en.wikipedia.org/wiki/Dust

http://en.wikipedia.org/wiki/Gas_bearing



Additionally, because there are no mechanical components, EFDs can require up to 98 percent less energy per IOPS

than traditional disk drives.

Until now, enterprises with high performance requirements had limited choices. They could either take a costly

approach of spreading workloads over dozens or hundreds of underutilized disk drives, or they could purchase

separate and expensive servers and memory storage that can add complexity and create storage islands. Now, using

EFDs, Tier 0 applications can be closely coupled with other storage tiers within Dell/EMC storage arrays for

consistency and efficiency, helping to eliminate the need for time invested in manual data layout or end-of-day data

transfers from separate RAM disk or specialized memory storage systems.

Performance of 73 GB EFDs

Enterprise flash drives do not contain any moving parts within the drive. This, combined with the multiple parallel

I/O channels on the back end, allows EFDs to show performance characteristics that are multiples above a

traditional rotating drive. EFDs show the highest improvement in performance (compared to the best performing

rotating drives) when the drive workload has a high percentage of small, random read operations and the application

is multi-threaded.

The typical expectation for an FC device is about 180 IOPS; the typical expectation for an EFD device is about

2,500 IOPS. Note that these throughputs values are measured at the drives; when measuring throughputs at the

host/server, parity calculations should also be included. In carefully tuned multi-threaded random small-block

environments, EFDs can deliver much lower response times and higher throughput than rotating drives. However,

production workloads that currently run on FC drives should not necessarily be expected to scale at the same

multiplier as drive IOPS since replacing the disk drive technology does not change the application thread count or

host/SAN service times.

Like their rotating counterparts, EFDs have a wide range of operating points that depend on the nature of the

workload that they service. For large block I/Os (64 KB and higher), EFDs tend to use all internal I/O channels in

parallel. Also, since single-threaded sequential I/O streams on FC drives do not suffer seek and rotational latencies

(because of the array cache), single-threaded large-block sequential I/O streams will not show major performance

improvements with EFD over FC drives. However, with increased application concurrency (as more threads are

added), the load starts to resemble a large block-random workload. In this case, seek and rotational latencies are

introduced that decrease FC drive effectiveness but do not decrease EFD effectiveness.

Applications using 73 GB EFDs

This section describes some of the applications in which you can use the new EFDs to help realize true performance

benefits. Based on the performance considerations mentioned in the previous sections, some general areas of interest

for EFD deployment are as follows:

Hot database tables that require very fast read response time and/or high throughput

Database temp area

High write bandwidth applications, like DB load and system backup, that are known to cause forced flushing

Small, highly active file systems that have a response time SLA

Metadata control areas in clustered file systems that require minimum read response time (and not necessarily

high throughput)

Typical requirements of applications selected for EFD deployment are:

A minimum small block read response time

The ability to scale throughput linearly over many concurrent threads

Avoidance of forced flushing.

Detailed performance guidance for using enterprise flash drives can be found in Dell/EMC Performance and

Availability: Release 28 Firmware Update – Applied Best Practices available on www.dell.com.

http://www.dell.com/



Fibre Channel Hard Drives

Dell/EMC 146, 300, and 450 GB 15k rpm (4 Gb/s) FC drives The Dell/EMC 146, 300, and 450 GB 15k rpm 4 Gb/s hard disk drives can dramatically increase performance over

previous HDDs through improvements in disk operations such as rotational latency and seek rates—the factors that

most directly affect access times. These disk drives deliver 3.5 ms average read seek times and 2.0 ms average

rotational latency times—both some of the fastest available in today's hard disk drive market.

Details about other capacity 15k rpm (4 Gb/s) FC drives supported on Dell/EMC systems can be found in the CX4

Series Storage Systems Disk and FLARE OE Matrix available on EMC Powerlink.

The following section illustrates how the increase in spindle motor rotation speed can make a difference in overall

I/O performance when compared to slower rotational-speed hard disk drives.

Performance of 15K RPM Disk Drives

Analysis of 15k rpm disk drive performance shows that not much of a performance benefit is realized over slower

spindle (for example, 10k rpm) drives when running applications that are sequential, read I/O intensive. This is due

to the fact that when running sequential I/O, there is virtually no head carriage movement, and spindle speed has

little effect in the overall access to sequential data coming from the media surface. Even with larger block sizes in

the equation, there is no discernible difference in the overall transfer rates between 10k and 15k rpm disk drives.

The overall performance difference between a 15k rpm and 10k rpm disk drive increases when I/O types are varied,

such as random read and random write operations. Larger performance improvements are realized when running the

15k rpm disk drives in random I/O environments. Drives of 15k rpm can yield up to a 35 percent performance

improvement in a random environment when compared to the 10k rpm drives.

Capacity does not have an appreciable effect on the performance of disk drives. For example, there is not much

performance difference between a Seagate 146 and a 300 GB 15k rpm disk drive.

Applications Using 146, 300, and 450 GB 15K RPM FC Drives

This section describes some of the applications in which you can use the 15k rpm disk drives to help realize true

performance benefits. Using the 15k rpm drives in applications that use small block, random I/O is an important

factor in realizing higher performance benefits. These applications have a tendency to minimize any caching

advantages of the storage system. In addition, with applications such as these, the physical access to data on the disk

has the greatest effect on overall performance. These small block, random I/O applications reap the greatest benefits

from storage-system performance improvements achieved through the new drive’s improved seek and rotational

latency times.

Some of the more popular types of applications for the 15k rpm drives include:

OLTP

E-commerce

ERP

Database

Web server

E-mail

Data replication

Dell/EMC 300 and 400 GB 10K RPM (4 Gb/s) FC Drives The Dell/EMC 300 and 400 GB 10k rpm 4 Gb/s hard disk drives can dramatically increase capacity and can lower

power consumption over their 2 Gb/s counterparts through improvements in disk operations such as spindle motor

technology, interface chip sets, and disk platter densities—the factors that most directly affect storage capacities and

power consumption. These disk drives deliver 3.9 ms average read seek times and 2.98 ms average rotational

latency times.



Details about other capacity 10k rpm (4 Gb/s) FC drives supported on Dell/EMC systems can be found in the CX4

Series Storage Systems Disk and FLARE OE Matrix available on EMC Powerlink.

Several interesting points to note are:

Rotational Speed

The 300 GB drives spin at 10k as do the 146 GB drives.

Average Seek Times

The averages are virtually identical on the 300 GB drives when compared to the 146 GB drives.

BPI and the Internal Transfer Rate

The specifications on the 300 GB seventh generation drives are actually slightly higher than those of the sixth

generation Seagate 300 GB hard disk drives.

Dell/EMC CX4 series arrays are not sold with 2 Gb/s drives. However, they are still supported in CX4 and legacy

arrays as an upgrade. Details about 10k rpm (2 Gb/s) Fibre Channel drives supported on Dell/EMC systems can be

found in the CX4 Series Storage Systems Disk and FLARE OE Matrix available on EMC Powerlink.

Performance of 10K RPM Disk Drives

Looking at these specifications and taking into account the bits per inch (BPI) and internal transfer rates of the two

drives, several performance characteristics of the 300 GB drives become apparent. When comparing the same

number of spindles in a random read/write environment, the performance of the 300 GB hard disk drives and the

146 GB hard disk drives is virtually equal. This is because their performance specifications are almost identical.

Replacing 146 GB drives with an equal number of 300 GB drives, without increasing the amount of data, can greatly

improve performance because of the reduced seek distances. This advantage is reduced as new data is added to the

configuration.

If 146 GB drives are replaced by a lesser number of 300 GB drives, the performance is reduced, since the

throughput of the system is directly proportional to the number of spindles.

The performance of the 300 GB and 146 GB Fibre Channel hard disk drives are similar when applications are

sequential read and write I/O intensive. When running sequential I/O, there is virtually little or no head carriage

movement, and the overall increase in BPI helps to maintain similar performance characteristics of the 146 GB and

the 300 GB disk drives. Even with larger block sizes, there is still no discernible difference in the overall transfer

rates between 146 and 300 GB 10k rpm disk drives.

To summarize, when using the 300 GB drives in the right environment, there is up to a 10 to 12 percent performance

improvement over the 146 GB previous-generation disk drives. In a sequential workload environment, the

performance implications of replacing 146 GB drives with 300 GB ones are different. Seek times are insignificant,

so there is no advantage to having more space. The spindle count change also has a lesser impact on throughput

because it takes fewer drives to limit system performance. In many cases, we can retain adequate throughput

performance in a bandwidth environment by using fewer spindles.

Applications Using 300 and 400 GB 10K RPM Disk Drives

To realize cost and capacity benefits, it is important to use the new 300/400 GB 10k rpm drives in applications

suited to higher-capacity environments. Sequential access applications have a tendency to maximize any caching

advantages of the storage system, and take advantage of the increased internal transfer rate of the drive. In these

applications, the speed at which the drive can transfer data from the platter has the greatest effect on overall

performance. Thus, medium-to-large block and sequential I/O applications can reap the greatest benefits from the

new drive’s improved internal transfer rates and higher bit densities.

Some of the more popular types of applications for the 300/400 GB 10k rpm drives are:

Database environments

Online backup for Internet services

Near-line storage or tape replacement

Oil and gas exploration

Life sciences



Digital A/V and digital editing

Medical imaging

Image archival

Document management

Data warehousing and data mining

The customer benefits from 300/400 GB disk drive technology in these areas:

Price per megabyte

Reduced footprint

Reduced power and cooling requirements

Improved MTBF rates through drive consolidation

Available on the current CX series product lines

Serial Attach SCSI Hard Drives Serial Attach SCSI (SAS) is a computer bus technology primarily designed to transfer data to and from devices like

hard drives, CD-ROM drives, and so forth. SAS is a serial communication protocol for direct attached storage

(DAS) devices. It is designed for the corporate and enterprise market as a replacement for parallel SCSI, allowing

for much higher speed data transfers than previously available, and is backward-compatible with SATA drives.

(SATA drives may be connected to SAS controllers. However, SAS drives may not be connected to SATA

controllers.) Though SAS uses serial communication instead of the parallel method found in traditional SCSI

devices, it still uses SCSI commands for interacting with SAS end devices. Due to the lower loops speed of 3 Gb/s,

SAS drives are currently supported only on the Dell AX series arrays.

Dell AX4-5 400 GB 10K and 15K RPM SAS Disk Drives SAS is the next generation of the Small Computer System Interface. Parallel SCSI has been the standard server and

workstation internal disk storage interface for over 20 years, and was also the standard device interconnect for open

systems external array storage prior to the advent of native Fibre Channel drives.

The entry external storage market, which has traditionally utilized either parallel SCSI or SATA drive technology,

has been the fastest adopter of SAS drive technology. For these systems, the ability to mix plug-compatible SAS

and SATA drives is a distinct advantage over parallel SCSI, which lacks tiered storage flexibility.

The performance difference between 10k and 15k rpm SAS disk drives is similar to performance difference between

10k and 15k rpm Fibre Channel drives described previously. Details about other capacity SAS drives supported on

Dell/EMC systems can be found in the CX4 Series Storage Systems Disk and FLARE OE Matrix.

Serial ATA Hard Drives

ATA has traditionally been used for internal storage interconnect in desktop computers to connect the host systems

to hard drives and optical drives. Today, the ATA interconnect technology has evolved for much higher interconnect

speeds, scalability, and reliability, surpassing the technology’s originally intended applications. ATA technologies

are now extensively used in enterprise class storage and server environments in near-line storage applications where

scale and costs are primary selection driving criteria. Serial ATA II is the next-generation internal storage

interconnect, designed to replace earlier ATA technologies (SATA I). This interconnect technology is capable of

communicating at speeds of 300 MB/s and is the technology of choice used in Dell/EMC storage systems.

Dell/EMC 1 TB 7200 RPM SATA II Drives Now we will look at some of the applications where we can implement 1 TB 7,200 rpm SATA II drives. Use SATA

II drives in applications that best suit higher-capacity environments to help realize cost and capacity benefits.

Sequential access applications have a tendency to maximize any caching advantages of the storage system, and take



advantage of the higher density SATA II disk drives. In these applications, the speed at which the drive can transfer

data from the platter has the greatest effect on overall performance. Thus, medium-to-large block and sequential I/O

applications can reap the greatest benefits from the drive’s higher area-bit densities.

Some popular types of applications for the 1 TB 7,200 rpm SATA II drives are:

Disk-to-disk backup

Disk backup using traditional backup software. Dell/EMC arrays with ATA is tested and supported with

most major backup applications.

Improves backup and restore performance when compared to tape.

Large application datasets

Some applications, like seismic data interpretation, government intelligence, and life sciences research are

immediately written out to tape due to their large size. When the tests need to be rerun, the data must be

reloaded from tape, and then rerun. Now the information can stay online with Dell/EMC SATA drives and

businesses can improve their operational efficiency and time to market.

Details about other capacity SATA II drives supported on Dell/EMC systems can be found in the CX4 Series

Storage Systems Disk and FLARE OE Matrix.

Benefits of 1 TB SATA II Disk Drive Technology

Lower price per megabyte

Reduced footprint

Reduced power and cooling requirements

Dell/EMC 1 TB 5400 RPM SATA II Drives (Lower Power SATA) Dell introduced a lower spindle speed 1 TB SATA II drive that delivers the highest density at the lowest cost and

can use 96 percent less energy per terabyte than the 73 GB 15k rpm FC drives and 32 percent less than traditional 1

TB 7,200 rpm SATA II drives1. This drive has been designed to maximize power savings for large, high-capacity

deployments such as array-based backup to disk, online tape replacement, and data warehousing. It is ideal for

applications where low power usage and cost-to-performance ratio is important and high performance is not a

priority.

The drives are available as 15-drive bundles pre-populated in a CX4-4PDAE for new systems, or as upgrades to

CX3 and CX4 systems. Installation in the same enclosure with other Fibre Channel or SATA drives model is not

supported.

Performance difference between 7,200 rpm and 5,400 rpm 1 TB SATA II drives will depend on the actual

application as well as the system configuration, including the number of drives and the RAID type deployed. In

highly sequential, large-block environments such as a lisk library, performance of the 5,400 rpm drives has been

shown to be comparable to the 7,200 rpm versions of 1 TB SATA II drives.

Detailed analysis of performance characteristics and differences between 1 TB 5,400 rpm SATA II drives and 1 TB

7,200 rpm SATA II drives can be found in the Dell/EMC Performance and Availability: Release 28 Firmware

Update – Applied Best Practices white paper available on www.dell.com .

Implementing 1 TB SATA II Disk Drives

Dell recommends that SATA drives be used for single threaded, large block streaming applications. In a typical

binding operation, various RAID groups and LUNs are bound and presented to the host as logical disk drives.

Within this binding/assignment operation, the hard disk drives are selected and grouped into usable storage for host

applications.

It is common practice to mix drive types and enclosures in Dell/EMC storage systems, according to user

requirements. This is where the 1 TB drives may be factored into the data capacity/performance mix. With the

1 Based on drive specifications. Actual power consumption will vary based on configuration, usage, and

manufacturing variability



current capacity points of the Dell/EMC disk drives at 73 GB EFD, 146/300 GB (10k rpm and 15k rpm), 400 GB

(10k rpm), and now the 1 TB 5400/7200 rpm SATA II disk modules, you can apply these different capacity and

performance drives/enclosures to suit the various applications within your operating environment.

SATA II "Northstar" Disk Drive and Enclosure Technology The Dell UltraPoint™ disk drive enclosure is the current generation Dell/EMC disk-array enclosure (DAE) that

replaces legacy 2 Gb/s FC loop technology DAEs that were used in the Dell/EMC 2 Gb/s CX300/CX500/CX700

series product line. The current UltraPoint design supports Fibre Channel Arbitrated Loop (FC-AL) interconnect

speeds of 4 Gb/s between the storage system and/or other DAE enclosures. The FC loop within the DAE is

implemented with point-to-point technology. This technology:

Emulates a loop for FC control traffic (loop primitives, initialization, and so forth).

Provides a point-to-point connection to each drive for data traffic. This feature provides improved isolation of

data traffic error conditions while reducing the loop latency of the FC data.

The UltraPoint storage solution consists of the following field-replaceable units:

3U rack-mountable DAE (UltraPoint enclosure) with a midplane and a 15-disk drive module capacity

Hot-swappable 1-inch low-profile FC disk modules

Hot-swappable link controller cards (LCCs)

Hot-swappable blower/power supplies

The SATA II disk drives in this enclosure incorporate a SATA II to FC bridge "paddle card" attached to the back of

the Dell/EMC SATA II disk drives. This bridge card resides on the disk drive module assembly, and combined with

a SATA II drive, emulates a single FC drive.

The main benefit of this technology is that you no longer need to separate ―ATA‖ type enclosures within the

Dell/EMC storage, because the new ―Northstar‖ SATA II disk drives can be incorporated with standard UltraPoint-

style Fibre Channel disk drive enclosures. This allows the Dell/EMC back-end loops to continue to run at 4 Gb/s.

Note: SATA II Northstar disk drives cannot be mixed with standard EFD/FC disk drives in the same enclosure.

The implementation of a Dell/EMC ATA enclosure is transparent to Dell/EMC software. Note, however, that there

are some limitations when implementing these ATA enclosures into the Dell/EMC CX series systems. The

limitations are:

RAID groups bound on ATA drives cannot span outside ATA drive enclosures, but can span through ATA

enclosures. In other words, ATA drives cannot be bound with FC drives.

Hot spares for ATA drives must be located in ATA enclosures and cannot be used as spares for FC disk

modules. (We recommend one spare ATA drive for every 30 ATA drives.)

ATA drives cannot be located within the first enclosure of the CX series storage system.

ATA drives should not be used for host booting activities due to performance concerns.

Random I/O environments should be reconsidered prior to implementing ATA drives in place of FC drives.

Other than these limitations, ATA drives may be in any configuration.

When planning for performance and capacity in an environment, the performance, pricing, and capacity

requirements for each application should be completely understood before selecting the drive(s). Select the 1 TB

SATA II drives when they meet both the storage capacity and performance requirements the individual applications

demand.

Disk Drive Performance Comparisons The following points should be noted about different drive types:

Rotational speed — The SATA II drives spin at 5,400 or 7,200 rpm, whereas the FC/SAS drives spin at 10,000

or 15,000 rpm. EFDs do not have any moving parts.



Since EFDs do not have any moving parts, they have no rotational and seek latencies.

Average seek times — The averages are quite a bit slower on the SATA II drives when compared to the FC

drives.

When you take into account the bits per inch (BPI) and internal transfer rates of the two drives, several performance

characteristics of the 1 TB SATA II drives become apparent. With average read and write seek rates of 9 ms to 10

ms, and an average rotational latency specification of 4.1 ms, the SATA II drives may not be well suited for random

I/O environments, such as database or OLTP environments. Instead, the Dell/EMC SATA II drives are ideal for

bringing offline information online. Offline applications use large, sequential-type data access and storage activities.

Some of these applications are discussed later in this paper.

Competitive Advantages of the Dell/EMC with ATA Following are some of the competitive advantages to implementing Dell/EMC SATA II drives into a new or existing

Dell/EMC infrastructure. A Dell/EMC SATA II drive:

Plugs in to an existing sixth and seventh generation architecture

Offers the full software functionality of Dell/EMC with SATA II technology

Can nondisruptively expand capacity by adding SATA II technology to an existing Dell/EMC CX or CX3 series

storage system

Provides dual porting on each SATA II drive for high availability

Makes hot plug and hot swaps available for all Dell/EMC ATA components

Includes FLARE advantages, such as Data Integrity Checking and Scrubbing (Sniffer)

Provides redundant high-availability components (for example, power, cooling, LCCs)

Includes the same RAID support (0, 1, 1/0, 3, 5, and 6) and drive-selection flexibility as Fibre Channel drives

Uses the Dell/EMC Sector Data Protection Scheme

Provides a checksum architecture for end-to-end data protection

Implementing Various Technology Dell/EMC Disk Drives In a typical binding operation, various RAID groups and LUNs are bound and presented to the host as logical disk

drives. It is within this binding/assignment operation that the hard disk drives are selected and grouped into usable

storage for host applications.

It is common practice in the Dell/EMC storage systems to mix drive types according to end-user requirements. With

the current capacity points of the Dell/EMC hard disk drives at 73 GB (EFD), 146/300/450 GB (15k rpm), 400 GB

(10k rpm), and 1 TB (7200 rpm and 5400 rpm), you can apply these different capacity and performance drives to

suit the various applications within your operating environment.

A Few Examples of Mixed Disk Drive Usage Following are examples of mixed disk drive usage:

In a CAD/CAM environment, the logical choice of a hard disk drive is higher-capacity storage per drive—for

drawing retention and design change—as opposed to lower-capacity storage per spindle.

In a price-sensitive environment, the lower-capacity hard disk drives may be the right choice due to the lower

initial cost of ownership of storage in these types of applications.

In high I/O, small block applications, such as OLTP, EFD, or 15k rpm drives, are a good fit from a performance

standpoint.

If we take the Dell/EMC modular design into account, we could then accommodate all four scenarios into one

Dell/EMC storage system by simply installing:

A DAE or RAID group with 1 TB, 7200 rpm drives for our near-line disk backup-based applications.

A DAE or RAID group with 1 TB, 5400 rpm drives for our CAD/CAM environment for drawing retention.

A DAE or RAID group with 400 GB, 10k rpm drives for our engineering-group applications.



A DAE or RAID group of high-performance 73 GB EFD and 146 or 300 GB, 15k rpm drives for OLPT and

random database applications.

A DAE or RAID group with 400 GB 10k rpm disk drives for a good midpoint-cost/performance-effective

storage solution.

Dell/EMC Disk Drive Power Solutions Dell/EMC’s advanced storage technologies can help improve overall system performance and can help optimize

storage energy use. In the following sections we discuss the foundations of energy usage and Dell/EMC best

practices for improving energy efficiency.

Multi-Tiering Dell/EMC has the ability to make effective use of multiple tiers of storage capacities within the same system.

Dell/EMC’s UltraScale architecture allows a wide variety of disk drive technologies within the same array. A single

UltraPoint disk-array enclosure (DAE) can support multiple drives (either SATA or EFD, FC, and/or Low-Cost

Fibre Channel) and multiple interface speeds (2 Gb/s and/or 4 Gb/s). The flexibility and configuration options

provided by use of UltraPoint technology, in conjunction with Virtual LUNs, enable you to easily move between

any of these drive technologies.

It is common practice in the Dell/EMC storage systems to mix drive types according to end-user requirements. This

is where 1 TB 5400/7200 rpm SATA II drives should be factored into the data capacity/performance mix. With the

current capacity points of the Dell/EMC hard disk drives at 73 GB (EFD), 146 GB, 300 GB and 450 GB (15k rpm),

400 GB (10k rpm) and 1 TB (5400/7200 rpm), you can select the best drive to suit the capacity and performance

requirements of each application in your environment. Figure 3 shows the power savings with different capacity

versions of FC and SATA II drives.

Figure 3 Power savings with different capacity FC and SATA II drives for 1 TB raw capacity2

The ability to move data dynamically from one tier to another in the Dell/EMC storage array can result in significant

energy savings as different storage tiers have a different power profiles. Table 2 shows the power consumption

2 Based on drive specifications. Actual power consumption will vary based on configuration, usage, and

manufacturing variability



profile of different types of disks that are available in Dell/EMC. Note that EFDs have very little difference between

active and idle power consumption because of the absence of any moving parts in the drive.

Table 2. Power usage between active and idle drives

Rotation speed

Active power Idle power Difference between idle and active power

15k rpm 24.15 W 20.99 W 13%

10k rpm 23.35 W 18.91 W 19%

7.2k rpm 21.35 W 11.51 W 46%

5.4k rpm 16.98 W 10.57 W 38%

EFD 15.02 W 15.02 W 0%

It is crucial to consider several factors when planning for performance, capacity, and energy requirements for any

given environment. Select the proper disk drive drives that meet the storage capacity, price point, power

consumption, and performance requirements that your individual applications demand. Table 3 details typical power

consumption when comparing different disk drive technologies in a standard Dell/EMC environment. Note that

these figures compare only the drive technology choice and not their relative energy per capacity.

Table 3. Typical power consumption of different drive technologies

Drive type Number of drives

Line current Power consumption

Annual energy costs

Heat dissipation

15k rpm 15 1.83 A 0.381 kVA $976 1,240 Btu/Hr

10k rom 15 1.77 A 0.368 kVA $943 1,200 Btu/Hr

7.2k rpm 15 1.62 A 0.337 kVA $863 1,100 Btu/Hr

5.4k rpm 15 1.29 A 0.268 kVA $686 870 Btu/Hr

EFD 15 1.14 A 0.237 kVA $607 770 Btu/Hr

Enclosure power consumption calculations based on a local utility rate of 0.1537 $ /kW-hr

Using one disk drive type for all load types on a storage system is not a recommended practice, although some

vendors may propose this practice. Each storage tier has a different information and power requirement throughout

its lifecycle. Thus, the different tiers of storage should be a major consideration when deciding on system

configuration/layout for storing and managing data.

Rebuild Times

Rebuild times depend on a number of factors, including the rebuild rate (ASAP, High, Medium, or Low),

presence/location of an appropriate hot spare, disk type, disk size, bus speed, application load, and RAID group

topology. When a hot spare is used for the rebuild, there is an additional ―equalize‖ operation that occurs when the

faulty drive is replaced and the hot spare is copied to it.

Equalize rates on RAID 5 are faster than the rebuild itself since only two drives are active (the hot spare and the new

drive) and there is no need to recompute parity or data from the other drives. Equalize on RAID 1/0 is identical to

the rebuild, only in the opposite direction. The effect of ASAP rebuild on application degradation depends on the

workload mix. If an ASAP rebuild cannot be tolerated by an application, rebuild time can be paced with the High,

Medium, or Low setting. The rebuild rates are much slower than ASAP, but at the High setting production

workloads are competing with the rebuild only for 10 percent of the time.

Baseline ASAP rebuild/equalize rates for common RAID group configurations using 15k rpm FC drives on a single

4 Gb/s loop can be found in the Dell/EMC Performance and Availability: Release 28 Firmware Update – Applied



Best Practices white paper. Information about performance degradation for different priority settings during a

rebuild process is also available in this paper.

Performance Planning

Performance planning or forecasting is a science that takes considerable knowledge. The steps presented here are

intended for rough estimation only.

Rule-of-thumb approach To begin performance estimation, a rule of thumb is used for IOPS per disk drive and MB/s per disk drive. This is a

conservative and intentionally simplistic measure. It should be noted that this is only the beginning of an accurate

performance estimate; estimates based on the rule of thumb are for quickly sizing a design. More accurate methods

are available to Dell personnel.

The approach for a quick estimate is:

Determine host IOPS or bandwidth load.

Calculate disk IOPS or bandwidth load.

Calculate the number of disk drives required for disk IOPS or bandwidth load.

Calculate the number and type of storage systems.

Rule-of-thumb numbers for all the Dell/EMC drive technologies and speeds are explained in Dell/EMC

Performance and Availability: Release 28 Firmware Update – Applied Best Practices.

Dell Hard Drive Reliability Qualification In general, EFD, FC, and SAS hard drives have ―Enterprise‖ level reliability. SATA drives have near-line

reliability. Prior to releasing any new technology, such as EFD and SAS, several technology compatibility and

characterization tests are conducted. In all cases, the drives are subjected to extreme environmental stresses and a

great deal of performance parameters are collected.

Following is a summary of the steps Dell takes to move a disk drive from definition to manufacturing and a

customer readiness phase for general availability.

Dell and EMC have established a world-class disk qualification and reliability process to help ensure adherence to

the highest standards for our customers. Only after going through these exhaustive lists of processes and

qualification processes is a drive then sent into manufacturing.



Conclusion Dell provides several different disk drive technologies for various applications, capacities, cost points, and use cases.

As a result of the varying capacities and performance characteristics currently supported on the Dell/EMC CX4

storage systems, we are now able to support four types of applications—meeting four different price/performance

tradeoffs within one Dell/EMC storage system platform.

We strive to deliver highest standards in overall system performance, scalability, and reliability.

An Introduction to EMC CLARiiON Storage Device Technology

Documents

Transcript of An Introduction to EMC CLARiiON Storage Device Technology